ERBB2 Induces an Antiapoptotic Expression Pattern of Bcl-2 Family Members in Node-Negative Breast Cancer

Published OnlineFirst January 12, 2010; DOI: 10.1158/1078-0432.CCR-09-1617 Published Online First on January 12, 2010 as 10.1158/1078-0432.CCR-09-1617

Clinical Human Cancer Biology Cancer Research ERBB2 Induces an Antiapoptotic Expression Pattern of Bcl-2 Family Members in Node-Negative Breast Cancer

Ilka Brigitte Petry1, Esther Fieber1, Marcus Schmidt1, Mathias Gehrmann2, Susanne Gebhard1, Matthias Hermes3, Wiebke Schormann3, Silvia Selinski3,4, Evgenia Freis3,4, Holger Schwender4, Marc Brulport3, Katja Ickstadt4, Jörg Rahnenführer4, Lindsey Maccoux5, Jonathan West5, Heinz Kölbl1, Martin Schuler6, and Jan Georg Hengstler3

Abstract Purpose: Members of the Bcl-2 family act as master regulators of mitochondrial homeostasis and apoptosis. We analyzed whether ERBB2 influences the prognosis of breast cancer by influencing the proapoptotic versus antiapoptotic balance of Bcl-2 family members. Experimental Design: ERBB2-regulated Bcl-2 family members were identified by inducible expression of ERBB2 in MCF-7 breast cancer cells and by correlation analysis with ERBB2 expression in breast carcinomas. The prognostic relevance of ERBB2-regulated and all additional Bcl-2 family members was determined in 782 patients with untreated node-negative breast cancer. The biological relevance of ERBB2-induced inhibition of apoptosis was validated in a murine tumor model allowing conditional ERBB2 expression. Results: ERBB2 caused an antiapoptotic phenotype by upregulation of MCL-1, TEGT, BAG1, BNIP1, and BECN1 as well as downregulation of BAX, BMF, BNIPL, CLU, and BCL2L13. Upregulation of the antiapoptotic MCL-1 [P = 0.001, hazard ratio (HR) 1.5] and BNIP3 (P = 0.024; HR, 1.4) was associated with worse prognosis considering metastasis-free interval, whereas clusterin (P = 0.008; HR, 0.88) and the proapoptotic BCL2L13 (P = 0.019; HR, 0.45) were associated with better prognosis. This indicates that ERBB2 alters the expression of Bcl-2 family members in a way that leads to adverse prognosis. Analysis of apoptosis and tumor remission in a murine tumor model confirmed that the prototypic Bcl-2 family

member Bcl-xL could partially substitute for ERBB2 to antagonize tumor remission. Conclusions: Our results support the concept that ERBB2 influences the expression of Bcl-2 family members to induce an antiapoptotic phenotype. Antagonization of antiapoptotic Bcl-2 family members might improve breast cancer therapy, whereby MCL-1 and BNIP3 represent promising targets. Clin Cancer Res; 16(2); 451–60. ©2010 AACR.

ERBB2 overexpression or amplification is frequently, represent a more effective breast cancer therapy compared although not invariably, associated with poor clinical with targeting only the individual factors. outcome (1–5). Recently, ERBB2 gene amplification Only little is known about the relationship between has been reported to correlate with the expression of ERBB2 and Bcl-2 family members. Members of the Bcl-2 apoptosis-suppressing genes Bcl-2 and Bcl-xL in breast family act as master regulators of mitochondrial homeo- cancer (6). Downregulation of Bcl-2 and Bcl-xL with anti- stasis and apoptosis (8, 9). Transfection of ERBB2 into sense oligonucleotides increased the proapoptotic activity MCF-7 breast cancer cells has been shown to upregulate of trastuzumab (7). This led to the hypothesis that the expression of Bcl-2 family members (10, 11). On the combined targeting of ERBB2 and Bcl-2 or Bcl-xL might other hand, antagonization of ERBB2 by trastuzumab has

Authors' Affiliations:1Department of Obstetrics and Gynecology, M. Schuler and J.G. Hengstler share senior authorship. University Medical Center of the Johannes Gutenberg University, Corresponding Author: Ilka B. Petry, Department of Obstetrics and Gy- Mainz, Germany; 2Siemens Medical Solutions, Diagnostics GmbH, necology, University Medical Center of the Johannes Gutenberg Univer- Cologne, Germany; 3 Leibniz Research Centre for Working sity, Mainz, Germany. Phone: 49-6131-177049; Fax: 49-6131-175673; Environment and Human Factors (IfADo), and 4Faculty of Statistics, E-mail: [email protected] and Jan G. Hengstler, Leibniz Research TU Dortmund University, 5ISAS-Institute for Analytical Sciences, Centre for Working Environment and Human Factors (IfADo) at the TU Dortmund, Germany; and 6Department of Medicine (Cancer Research), Dortmund University, Ardeystraâe 67, 44139 Dortmund, Germany. West German Cancer Center, University Hospital Essen, Essen, Phone: 49-231-1084-348; E-mail: [email protected]. Germany doi: 10.1158/1078-0432.CCR-09-1617 Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/). ©2010 American Association for Cancer Research.

www.aacrjournals.org 451

Petry et al.

by Baasner et al. (17), resulting in a cell line termed Translational Relevance NIH3T3-HER2. Cells with conditional ERBB2 expression and constitutive In this article, we identified ERBB2-dependent fac- BclxL expression. The pM_bcl-xl_pie (MS87) vector (Sup- tors of the Bcl2 family by inducibly overexpressing plementary Fig. S1) was transfected into NIH3T3-HER2 ERBB2 in a breast cancer cell line and by comparing cells resulting in a cell line named NIH3T3-HER2-BclxL these factors in ERBB2 status negative and positive constitutively expressing BclxL. The antiapoptotic cDNA breast carcinomas. The prognostic relevance of the BclxL was subcloned into the vector pMxIG (18), generat- identified ERBB2-dependent Bcl2 family members ing the pM_Bcl-xl_pie vector and the insert was confirmed was analyzed in 782 patients with node-negative breast by sequencing (19). The same culture conditions were cancer. From a clinical standpoint, the association of used as for the NIH3T3-HER2 cells. The replication-defec- high clusterin (downregulated by ERBB2) expression tive recombinant retrovirus vector pMxIG expresses green with lower hazard ratio for disease-free survival and fluorescent protein (18) to be able to monitor transduc- time without metastasis is the most provoking result tion efficacies by parallel experiments with flow cytometric of the present study. This result is clinically relevant be- detection of green fluorescent protein expression (20). Be- cause phase I and II studies of agents targeting CLU are sides the pMxIG vector containing the BclxL construct, under way. The rationale is that clusterin is considered NIH3T3-HER2 cells were cotransfected with the vector to be a cytoprotective chaperone that promotes cell without cDNA resulting in a control cell line named survival and confers broad-spectrum treatment resis- NIH3T3-HER2-MOCK. Transfection and culture condi- tance. However, considering our data, CLU-antagoniz- tions were the same as for the NIH3T3-HER2-BclxL cells. ing therapy may also eliminate a beneficial factor for Mouse tumor model. NIH3T3-HER2 cells (7 × 106) or de- spontaneous tumor development. rivatives of them were s.c. injected into the dorsal skin of − − 3- to 4-wk-old male nude mice (cd nu /nu ; Charles River). The experimental design is shown in Fig. 1. A previously been reported to reduce the expression of the Bcl-2 family described mouse tumor model was applied in which member MCL-1 (12). Because an antiapoptotic cross-talk ERBB2 expression is controlled by the Tet-off system between ERBB2 and antiapoptotic Bcl-2 family members, (14). The expression of ERBB2 in these mouse tumors as suggested by Milella et al. (7), is of high clinical rele- was downregulated by s.c. injection of 10 mg of anhydro- vance, we addressed this topic using three different ap- tetracycline/kg body weight into the nude mice. For West- proaches. First, we used a MCF-7 breast cancer cell line ern blot analysis, tumors were treated for 1, 3, and 7 d conditionally expressing ERBB2 (13) and observed that with anhydrotetracycline. Thereafter, mice were sacrificed ERBB2 upregulated antiapoptotic and downregulated by cervical dislocation. Tumors were isolated as described proapoptotic Bcl-2 family members. Second, we analyzed previously (14). the clinical relevance of ERBB2-dependent Bcl-2 family Western blot analysis. Western blot analysis was done as members in a cohort of 782 patients with breast cancer previously detailed (2). Primary antibodies: Bcl-xL,cas- and observed an association with poor prognosis. Third, pase-3, and ERBB2 (Cell Signaling Technology, New Eng- we validated the clinical observations in a mouse tumor land Biolabs GmbH) were used at a dilution of 1:1,000 and β-actin (Sigma) at a dilution of 1:2,000. Secondary model (14) showing that antiapoptotic Bcl-xL partially substitutes for ERBB2 in survival signaling of experimental antibodies: rabbit anti-mouse IgG (Sigma) and goat anti- tumors in vivo. Taken together, these findings show the rabbit IgG (Cell Signaling Technology) were used at a di- clinical relevance of ERBB2-dependent modulation of lution of 1:5,000. Primary and secondary antibodies were apoptosis-regulating Bcl-2 family proteins. diluted in PBST containing 10% Roti-Block. Cultivation of MCF-7-NeuT cells and analysis of RNA expression patterns. The MCF-7 breast carcinoma cell line was Materials and Methods obtained from American Type Culture Collection, cultured at 37°C in a humidified 5% CO2 air atmosphere, and Cells with ERBB2 expression under the control of tTA. transfected with pINSpBI-NeuT/EGFP and pcDNA3Neo/ NIH3T3, an immortalized cell line originally derived from rtTA2 as described by Trost et al. (13). NeuT expression, mouse embryo fibroblasts, was obtained from American an oncogenic version of ERBB2, was induced by doxycy- Type Culture Collection. Wild-type NIH3T3 and its deriva- cline (obtained as the hydrochloride salt from Sigma) at tives were grown in DMEM (PAN) supplemented with a final concentration of 1 μg/mL in all experiments. The 10% fetal bovine serum (tetracycline free; Clontech), and generated MCF-7-NeuT cells were exposed to doxycycline 1% penicillin-streptomycin (PAN). Cells were cultured at for periods of 0, 6, 12, and 24 h as well as 3 and 14 d. 37°C in 5% CO2 humidified air. Conditional expression Three independent repeat experiments were done. Be- of ERBB2 was achieved by using the Tet-off system origi- tween each repeat experiment, the MCF-7-NeuT cells were nally described by Gossen et al. (15, 16). Briefly, wild-type cultivated for at least two passages. Cells were harvested NIH3T3 cells were cotransfected with three vectors (pUHD using TRItidy G-Reagent (AppliChem GmbH) and subse- 15-1, pTBC1 Hygro, and pTBC HER2/SEAP) as described quently RNA was isolated according to the protocols of the

452 Clin Cancer Res; 16(2) January 15, 2010 Clinical Cancer Research

Bcl-2 Family Members in Breast Cancer

manufacturerandstoredat−80°C. Before microarray were validated in two previously published microarray da- analysis, RNA integrity and concentration were examined ta sets. Two breast cancer Affymetrix HG-U133A microar- using an Agilent 2100 Bioanalyzer (Agilent Technologies). ray data sets including patient outcome information were We used the RNA 6000 LabChip Kit (Agilent Technolo- downloaded from the National Center for Biotechnology gies) according to the instructions of the manufacturer. Information GEO data repository. The first data set, the Microarray analysis was conducted at the microarray core Rotterdam cohort (22), represents 180 lymph node–nega- facility of the Interdisziplinäres Zentrum für klinische For- tive relapse-free patients (GSE2034) and 106 lymph schung, Leipzig, Germany (Faculty of Medicine, University node–negative patients that developed a distant metasta- of Leipzig). Immunoblot analysis was done as described sis. None of these patients had received systemic neoadju- previously (13, 21). vant or adjuvant therapy (Rotterdam cohort). The original Patient characteristics, tissue specimens, and gene array data were recalculated to a TGT of 500. The second data analysis. Three recently described cohorts of 782 patients set, the Transbig cohort, consists of 302 samples from pa- with node-negative breast cancer, who did not receive che- tients with breast cancer that remained untreated in the motherapy, were analyzed (1). Patient characteristics, in- adjuvant setting after surgery (23,24). GSM numbers of cluding the frequency of ERBB2, estrogen receptor (ER), samples (from GSE6532 and GSE7390) used for analysis as well as progesterone receptor (PR) have already been are listed in the Supplementary Tables previously pub- published by our group (1). The combined cohort con- lished by Schmidt et al. (1). Raw.cel file data were pro- sisted of three subcohorts, the Mainz (n = 194), the Rot- cessed by MAS 5.0 using a target value (TGT) of 500. terdam (n = 286) and the Transbig (n = 302) cohorts. The ERRB2 and ER as well as PR status of the Mainz cohort Mainz study cohort (1), the only cohort for which multi- were defined by immunostaining and in case of ERBB2 by variate analysis were done in this study, consisted of 200 combined immunostaining and fluorescence in situ hy- consecutive lymph node-negative breast cancer patients bridization as described below. Immunohistochemical treated at the Department of Obstetrics and Gynecology analyses were done on 4-μm-thick sections according to of the Johannes Gutenberg University Mainz between standard procedures. Serial sections of formalin-fixed 1988 and 1998 (1). Six borderline cases were not included and paraffin-embedded tumor tissues were stained with in the present study resulting in a case number of 194. Pa- monoclonal ER antibodies (clone 1D5, Dako), monoclo- tients in the Mainz cohort were all treated with surgery and nal PR antibodies (clone PgR 636, Dako), as well as poly- did not receive any systemic therapy in the adjuvant set- clonal ERBB2 antibodies (A0485, Dako). The sections ting. The established prognostic factors (histologic grade, were incubated with a biotin-labeled secondary antibody tumor size, and age at diagnosis) were collected from and streptavidin-peroxidase for 20 min each, then for 5 the original pathology reports of the gynecologic patholo- minwith0.05%3′3-diaminobenzidine-tetrahydrochlor- gy division within our department. Patients' characteristics ide, and finally counterstained with hematoxylin. All series have been published by Schmidt et al. (1). For all tumors, included appropriate positive and negative controls and samples were snap-frozen and stored at −80°C. Gene ex- all controls gave adequate results. ERBB2 was scored from pression profiling of the patients' RNA was done using the 0 to 3+ according to the instructions published by the Affymetrix HG-U133A array and the GeneChip System as manufacturer. All ERBB2 2+ cases were confirmed by fluo- described (1). Results obtained from the Mainz cohort rescence in situ hybridization using a dual-color probe

Fig. 1. Experimental design of the mouse tumor experiments. Nude mice received s.c. injections of NIH3T3 cells expressing ERBB2

(ERBB2), ERBB2 and Bcl-xL (ERBB2-Bcl-xL), as well as ERBB2 expressing cells mock transfected with an empty vector (ERBB2-MOCK). As soon as the individual tumors reached a volume of 1.3 cm3, ERBB2 was downregulated by the Tet-off system. Subsequently, the tumor volume was monitored (Fig. 3) and tumor tissue was analyzed by immunoblot (Fig. 2).

www.aacrjournals.org Clin Cancer Res; 16(2) January 15, 2010 453

Petry et al.

Table 1. Induced expression of NeuT, an oncogenic version of ERBB-2, causes predominantly antiapoptotic alterations in RNA expression of Bcl-2 family members and proteins interacting with Bcl-2 family members

Symbol (gene) Function Expression after induction of NeuT* (fold expression) 6 h 12 h 24 h 3 d 14 d

Bcl-2 family members MCL-1 (myeloid cell leukemia 1) Antiapoptotic protein of the Bcl-2 2.519 3.106 2.198 2.319 1.864 family that heterodimerizes with other Bcl-2 family members to protect against apoptotic cell death (37) BMF (BCL2-modifying factor) Interaction with prosurvival Bcl-2 −1.576 −1.590 −1.577 −1.315 −1.205 family members to trigger apoptosis (38) BCL2L13 (BCL2-like 13) Inhibits cell death upon growth −1.355 −1.321 −1.279 −1.133 −1.056 factor withdrawal similar to Bcl-2 (39) BNIP1 (BCL2/adenovirus E1B BNIP1 variants interact with 1.973 1.844 1.559 1.787 1.052 19-kDa protein-interacting Bcl-2 and also with Bcl-xL protein 1) (30, 33) BNIP3 (BCL2/adenovirus E1B Bcl-2 proapoptotic family member 1.023 1.202 1.669 1.017 1.043 19-kDa protein-interacting recently shown to induce necrosis protein 3) rather than apoptosis. Interacts with other Bcl-2 family members (31–33) BAX (BCL2-associated X protein) Induces apoptosis by binding to 1.056 −1.237 −1.204 −1.424 −1.586 the mitochondrial permeability transition pore complex (37, 40) Direct protein interaction with Bcl-2 family members BNIPL (BCL2/adenovirus E1B BCH domain of the BNIPL2 i −1.383 −2.062 −1.814 −1.685 −2.202 19-kDa protein-interacting soform interacts with Bcl-2, protein 2-like) proapoptotic function (41) TEGT (testis-enhanced gene Interactions between the proteins 0.012 0.068 −0.012 1.571 1.545 transcript) of the TEGT and the Bcl-2 families, stimulation of the antiapoptotic function of Bcl-2 or inhibition of the proapoptotic effect of BAX (42) CLU (clusterin) Inhibits apoptosis by interfering 1.050 1.257 −1.013 −4.095 −4.323 with BAX (26) BECN1 (beclin 1) Beclin interacts with Bcl-2 and 1.438 1.552 1.367 1.419 1.308 BCLXL but not with BAX (22) BAG1 (BCL2-associated Binds to Bcl-2, prevents 1.654 2.020 1.558 1.089 1.088 athanogene 1) fibroblasts from undergoing apoptosis (43, 44)

NOTE: NeuT was expressed in MCF-7 breast cancer cells in a doxycycline-dependent manner using the Tet-on system. RNA from MCF-7 cells was harvested 6, 12, and 24 h as well as 3 and 14 d after onset of exposure to doxycycline and analyzed by Affymetrix gene array. Three independent batches of MCF-7 cells were analyzed, in which the cells were cultured for at least two passages between the individual experiments and doxycycline incubations, as well as RNA isolations which were done at different days. Expression data are medians of the corresponding probe sets. We summarized all Bcl-2 family members and all factors directly interacting with Bcl family members that were upregulated or downregulated by a factor of at least 1.5 as a consequence of induced NeuT overexpression. *Expression of MCF-7 cells not exposed to doxycycline was used as a control and corresponds to a gene expression of 1.0.

454 Clin Cancer Res; 16(2) January 15, 2010 Clinical Cancer Research

Bcl-2 Family Members in Breast Cancer

(DakoCytomation) containing a spectrum orange–labeled hormone receptor status in positive (ER and/or PR posi- ERBB2 gene (17q11.2-q12) probe and a spectrum green– tive) and negative (both, ER and PR negative), histologic labeled centromere control for chromosome 17 (17p11.1- grade in GIII versus GI and GII, pT stage in pT2 and pT3 q11.1). Appropriate positive controls were included into (> 2 cm) versus pT1 (≤ 2 cm). Correlations were analyzed each staining run. Analysis was done with a Zeiss fluores- using the Spearman correlation test. All P values were cence microscope (Axioskop 2). A minimum of 80 non- two-sided. As no correction for multiple testing was overlapping nuclei were evaluated and the ratio of done, they should be regarded as descriptive measures. ERBB2 signals per nuclei relative to chromosome 17 cen- All analyses were done using SPSS17.0. tromere signals was calculated. Ratios of 2 and more were classified as ERBB2 amplification. ERBB2 2+ tumors with Results ERBB2 amplification were finally considered HER-2 positive. ER and PR were assessed using an immunoreactive ERBB2 expression induces an antiapoptotic phenotype in score defined by the product of a proportion score (0, breast cancer cells. To study the influence of ERBB2 ex- none; 1, <10%; 2, 10-50%; 3, 51-80%; 4, >80% positive pression on Bcl-2 family members, we used an MCF-7 cells) and an intensity score (0, no staining; 1, weak; 2, cell line (MCF-7-NeuT) that allows doxycycline-induced moderate; 3, strong). Only nuclear staining was consid- expression of NeuT, an oncogenic version of ERBB2 ered for ER and PR. Patients with a score of >1 were con- (13, 25). MCF-7-NeuT cells were exposed to doxycycline sidered positive for ER or PR, respectively. for 6, 12 and 24 hours as well as 3 and 14 days. RNA Survival analysis. Univariate and multivariate Cox mod- was harvested from three independent biological repli- els were applied to analyze a possible association be- cates and was used for Affymetrix gene array analysis. tween RNA expression of Bcl-2 family members and The number of genes altered by at least a factor of 1.5 prognosis. RNA expression levels of Bcl-2 family mem- after 6, 12, and 24 hours, 3 and 14 days following induc- bers were evaluated as continuous variables (log 2 trans- tion of NeuT amounted to 30, 103, 143, 208, and 223. formed data). Disease-free survival (DFS) was computed Among these genes were five members of the Bcl-2 fam- from the date of diagnosis to the date of local recurrence ily, i.e., MCL-1, BMF, BCL2L1, BNIP3, and BAX, as well of disease, distant metastasis, cancer of the contralateral as five direct protein interaction partners with Bcl-2 fam- breast, or death from cancer. The metastasis-free survival ily members (Table 1). The strongest induction was ob- interval (MFI) was computed from the date of diagnosis served for the antiapoptotic MCL-1,whereasthe to the date of diagnosis of distant metastasis. Patients proapoptotic factors BAX and BMF were markedly down- who died of an unrelated cause were censored at the date regulated. Further upregulated factors with antiapoptotic of death. Survival times were compared using Kaplan- functions were TEGT, BAG1, BNIP1, and BECN1,al- Meier plots and the log rank test. Dichotomization was though a role of the latter in apoptosis is still discussed done as follows: ERBB2 status in positive and negative, controversially (22). Further downregulated factors with

Table 2. Univariate Cox analysis of the prognostic influence of BNIP3, CLU, and BCL2L13 in three cohorts of untreated node-negative breast cancer

Bcl-2 family–associated Mainz* cohort Rotterdam cohort Transbig cohort Combined cohorts factor (n = 194) (n = 286)† (n = 302)† (n = 782)†

BNIP3 P value 0.048 0.024 0.814 0.082 HR 1.409 1.394 1.034 1.164 95% CI 1.004-1.978 1.044-1.861 0.781-1.370 0.981-1.381 CLU P value 0.001 0.006 0.721 0.008 HR 0.705 0.813 1.029 0.884 95% CI 0.569-0.873 0.703-0.941 0.880-1.202 0.806-0.969 BCL2L13 P value 0.019 0.049 0.161 0.265 HR 0.453 0.651 1.574 0.830 95% CI 0.235-0.876 0.424-0.998 0.835-2.964 0.598-1.152

NOTE: The three studied Bcl-2 family members amounted to statistical significance in two of three cohorts. Results with P < 0.1 values are italicized. Abbreviation: 95% CI, 95% confidence interval. *DFS was analyzed in the Mainz cohort. The respective data of MFI are given in Supplementary Table 6. †MFI is given for the Rotterdam and Transbig cohorts.

www.aacrjournals.org Clin Cancer Res; 16(2) January 15, 2010 455

Petry et al.

Fig. 2. Expression of Bcl-xL (A), ERBB2 (B), and active as well as inactive caspase-3 (C) in tumor tissue of nude mice after injection of the NIH3T3-ERBB2-Bcl-xL cells (Bcl-xL) or NIH3T3-ERBB2 (parental) cells (ERBB2). Tumor tissue was analyzed 3 d after onset of daily anhydrotetracycline injections (3d ATc) and was compared with time-matched control tumors (3d Co). Tumors resulting from injection of NIH3T3-ERBB2-Bcl-xL cells expressed high levels of Bcl-xL in contrast to the parental NIH3T3-ERBB2 cells (A). The extent of ERBB2 downregulation by anhydrotetracycline (ATc) was similar in both tumor types (B). Importantly, lower levels of active caspase-3 were observed in tumors expressing Bcl-xL (C). This was the case after anhydrotetracycline administration, but also in mice that did not receive anhydrotetracycline, showing that tumors from NIH3T3-ERBB2 cells have relatively high levels of spontaneous apoptosis.

possible proapoptotic functions are BNIPL and CLU analysis showed that high expression of BNIP3 was asso- (Table 1). With the exception of BNIP3, whose function ciated with worse prognosis in the Mainz and Rotterdam as a bona fide proapoptotic BH3 protein still remains to cohorts (Table 2). High expression of proapoptotic cluster- be genetically proven, all other genes fit into a pattern of in (CLU) was associated with better prognosis in the upregulation of antiapoptotic and downregulation of Mainz and Rotterdam cohorts and in the combined cohort proapoptotic factors. of 782 patients (Table 2). MCL-1 and BNIP1 were associ- Correlation of ERBB2-associated Bcl-2 family members ated with worse prognosis in the combined cohort of 782 with clinical outcome in node-negative breast cancer. It patients, but not in the individual cohorts (Supplementary should be considered that ERBB2-induced alterations Table S2). Multivariate analysis of the Mainz cohort re- in the expression of Bcl-2 family members were ob- vealed CLU as a prognostic factor independent from the tained in the breast cancer cell line MCF-7, which may established clinical variables age, pT stage, histologic not be representative for a population of breast cancer grade, ER and PR as well as ERBB2 status [P = 0.020; haz- patients. Therefore, we studied a possible correlation be- ard ratio (HR), 0.780; 95% confidence interval, 0.633- tween ERBB2 RNA expression and all Bcl-2 family mem- 0.961; Supplementary Table S3A]. BNIP3 expression was bers in the tumor tissue of 782 patients with breast associated with histologic grading (Supplementary Fig. cancer. Interestingly, four of five Bcl-2 family members S2) and, therefore, no independent prognostic factor (Sup- that were upregulated in MCF-7 cells after the induction plementary Table S3B). In summary, ERBB2 expression of ERBB2, i.e., TEGT, BAG1, BNIP1, and BECN1 (Table 1), correlates with a pattern of Bcl-2 family members that is were also positively correlated with ERBB2 in at least associated with worse prognosis. This includes downregu- one of the three cohorts of patients with breast cancer lated CLU, upregulated BNIP3, and upregulated MCL-1 (Supplementary Table S1). In addition, BAX was nega- and BNIP1 which are associated with worse prognosis in tively correlated with ERBB2 in the Transbig cohort ERBB2-positive and ERBB2-negative breast cancers. (P = 0.003), confirming the negative association between Role of additional Bcl-2 family members in breast cancer ERBB2 and BAX observed in MCF-7 cells. and their relation to ERBB2. In the previous paragraphs, Because 10 Bcl-2 family–related RNA species showed al- we have focused on Bcl-2 family members that were reg- tered expression levels after NeuT induction in MCF-7 ulated by ERBB2 in MCF-7 cells. Next, we systematically cells, we next set out to analyze the potential clinical rele- studied the prognostic relevance of all Bcl-2 family mem- vance of these factors. For this purpose, we used three co- bers (listed in Supplementary Table S4). However, apart horts of medically untreated patients with node-negative from the Bcl-2 family members already described in the breast cancer (n = 782) that have been analyzed by gene previous paragraphs, only one additional relevant factor, expression arrays: the Mainz cohort (1), the Rotterdam co- i.e., BCL2L13 could be identified (Table 2). High expres- hort (22), and the Transbig cohort (23, 24). Univariate sion of BCL2L13 was associated with a decreased HR for

456 Clin Cancer Res; 16(2) January 15, 2010 Clinical Cancer Research

Bcl-2 Family Members in Breast Cancer

MFI in the Mainz and in the Rotterdam cohorts. Due to Next, we studied whether Bcl-xL affects the tumor re- the lack of published clinical data in the Rotterdam co- sponse upon withdrawal of ERBB2 in vivo using the s.c. hort, BCL2L13 could be tested in the multivariate Cox injection of NIH3T3-ERBB2-Bcl-xL, NIH3T3-ERBB2, or model only for the Mainz patients. Here, it was a prog- NIH3T3-ERBB2-MOCK cells into nude mice. As soon nostic factor independent from the established clinical factors age, grading, pT stage, ERBB2, and hormone receptor status (Supplementary Tables S3C and S5 for DFS and MFI). A negative correlation between ERBB2 and BCL2L13 expression (P = 0.017) was found in the Mainz cohort (Sup- plementary Table S1). Therefore, BCL2L13 represents another Bcl-2 family member, which is possibly modified by ERBB2 in a way that may be associated with worse prognosis. Bcl-2 family member Bcl-xL partially substitutes for ERBB2 and antagonizes the success of proapoptotic ERBB2 blocking therapy. Our expression analysis suggests that the modulation of Bcl-2 family proteins is important for the biological function of the ERBB2 oncogene. To address this hypothesis at a functional level, we used a tumor model in which tetracycline-controlled downregulation of human ERBB2 causes apoptosis in tumor cells leading to a rapid tumor remission (2, 14, 17, 24). To dissect the role of apoptosis-regulating factor in ERBB2 signaling in vivo, antiapoptotic Bcl-xL was expressed in conditional NIH3T3-ERBB2 cells. Bcl-xL was chosen for this experiment because it represents one of the best-understood Bcl-2 family members acting similarly as MCL-1 and Bcl- 2. After injection of the (a) NIH3T3-ERBB2-Bcl-xL cells, (b) NIH3T3-ERBB2 (parental) cells, (c) NIH3T3-ERBB2- MOCK (expressing a control vector) into nude mice, all three cell lines formed rapidly proliferating tumors within 14 days (Experimental Design: Fig. 1). Tumors resulting from the injection of NIH3T3-ERBB2-Bcl-xL cells expressed high levels of Bcl-xL in contrast with the parental NIH3T3- ERBB2 and NIH3T3-MOCK cells (Fig. 2A; Supplementary Fig. S3A). Downregulation of ERBB2 by anhydrotetracycline was equally effective in all three cell types (Fig. 2B; Supplementary Fig. S3B), and did not influence the levels of Bcl-xL (Fig. 2A, Supplementary Fig. S3A).

Fig. 3. Bcl-xL expression in tumors antagonizes the therapeutic efficiency of ERBB2 downregulation. Subcutaneous tumors were induced in nude mice by injection of NIH3T3 cells expressing ERBB2, ERBB2 and

Bcl-xL, as well as cells expressing ERBB2 mock-transfected with an empty vector. Downregulation of ERBB2 was achieved by daily injections of anhydrotetracycline for 7 d. The alteration of tumor volume compared with the day of the first anhydrotetracycline injection (day 0) is given. A, NIH3T3 cells expressing ERBB2 and Bcl-xL (Bcl-xL) versus NIH3T3 cells expressing ERBB2 and mock-transfected with an empty vector (MOCK). B, NIH3T3 cells expressing ERBB2 and Bcl-xL (Bcl-xL) versus NIH3T3 cells expressing ERBB2 (ERBB2). C, NIH3T3 cells expressing ERBB2 and mock-transfected with an empty vector (MOCK) versus NIH3T3 cells expressing ERBB2 (ERBB2). Comparing mock-transfected cells to the parental NIH3T3-ERBB2 cells showed that the retroviral transduction itself might have induced a slight increase in resistance (C). However, this small difference cannot explain the much larger effect observed in the presence of Bcl-xL (A and B).

www.aacrjournals.org Clin Cancer Res; 16(2) January 15, 2010 457

Petry et al.

as the individual tumors reached a volume of 1.3 cm3, Having identified 10 ERBB2-associated Bcl-2 family mice received injections of anhydrotetracycline for 7 members, we next examined their prognostic relevance in subsequent days to downregulate ERBB2 transgenic patients with breast cancer. For this purpose, we used Affy- expression. As observed in previous studies (2, 14, metrix gene array data from 782 patients with node-nega- 17, 21), withdrawal of ERBB2 caused strong tumor tive breast cancer. Because established clinical prognostic remissions. Interestingly, remissions of tumors from variables such as age, pT stage, grading, hormone receptor NIH3T3-ERBB2-Bcl-xL cells were less pronounced as status, and ERBB2 status were only available in the Mainz compared with the NIH3T3-ERBB2-MOCK control tu- cohort, multivariate Cox analysis was only done in the lat- mors (Fig. 3A) and parental NIH3T3-ERBB2 (Fig. 3B) ter. High expression of proapoptotic clusterin (CLU), tumors. Previously, we have shown that apoptosis which has been shown to be negatively correlated with contributes to tumor remission after withdrawal of ERBB2 in MCF-7 cells, was associated with better prognosis ERBB2 (14, 21). Therefore, we studied cleaved (ac- in the combined cohort, and in the Mainz as well as the tive) caspase-3 in tumors after 3 days of anhydrote- Rotterdam cohorts. Multivariate analysis showed that tracycline administration (Fig. 2C). Higher levels of CLU expression is a prognostic factor independent from es- cleaved caspase-3 were observed in tumors from tablished clinical variables (Supplementary Table S3A). Be- NIH3T3-ERBB2cellsascomparedwithNIH3T3- cause CLU was downregulated in MCF-7 breast cancer cells ERBB2-BclxL cells. In conclusion, Bcl-xL counteracted in response to ERBB2 induction, this is one of several tumor remissions of an ERBB2-antagonizing treat- ERBB2-dependent Bcl-2 family members that correlate ment by antagonizing apoptosis in tumor cells. with adverse prognosis. From a clinical standpoint, the association of high clusterin expression with longer DFS and Discussion time without metastasis is the most provoking result of the present study. The association of CLU with better prognosis To better understand the relationship between ERBB2 was observed in the Mainz cohort (P = 0.001), the Rotter- and Bcl-2 family members, we first studied the extent to dam cohort (P = 0.006), and in the combined cohort of which the expression of Bcl-2 family members is influ- 782 patients (P = 0.008). Thus, from a statistical point of enced by conditional ERBB2 expression in the human view, the positive prognostic influence of CLU is very con- breast cancer cell line MCF-7, and subsequently, analyzed vincing. This result is clinically relevant because phase I and the prognostic relevance of the identified genes in patients II studies of agents targeting CLU are under way (27–29). with breast cancer. Inducible expression of NeuT, an onco- The rationale is that clusterin is considered to be a cytopro- genic version of ERBB2, was achieved by the Tet-on system tective chaperone that promotes cell survival and confers (13, 25), and ERBB2 expression levels were comparable broad-spectrum treatment resistance (27, 29). Indeed, sev- with breast tumors scoring 2+/3+. The induced ERBB2 eral studies have shown that silencing clusterin may en- was functional as evidenced by increased phosphorylation hance the cytotoxicity of several chemotherapeutic agents of ERK1/2, Akt, and P38 after induction of the Tet-on (for a review, see ref. 29). However, considering the pres- system by doxycycline. Time-dependent gene array analy- ence of two CLU isoforms, a nuclear form, that is proapop- sis was done ranging from 6 hours to 14 days after switch- totic, and a secretory form, that is prosurvival, and bearing ing on NeuT expression. NeuT expression induced an in mind that some cancers show decreased levels of secre- antiapoptotic expression pattern of Bcl-2 family members tory CLU (29), the role of clusterin in tumor development in which the antiapoptotic MCL-1, TEGT, BAG1, BNIP1, is not yet understood. Because our study includes a higher and BECN1 were reproducibly upregulated and four proa- number of cases (n = 782) as compared with previous stud- poptotic Bcl-2 family members and related genes were ies (n = 141 and n = 158; refs. 30, 31, respectively), the re- downregulated, i.e., BAX, BMF, BNIPL, and CLU (Table 1). ported trends towards worse prognosis for patients with Although the role of CLU in apoptosis is still discussed high CLU expression should be carefully re-evaluated. It controversially, overexpression has been reported to should be considered that CLU-antagonizing therapy induce cell death in tumor cells (26). In conclusion, ge- might, on the one hand, improve the sensitivity of tumor nome-wide analysis of RNA expression following the cells to cancer chemotherapy, but might also eliminate a induction of oncogenic ERBB2 transcription results in beneficial factor for spontaneous tumor development. In the expression of an antiapoptotic phenotype. our study, none of the patients had received chemotherapy, It should be considered that the association between thus focusing on the role of CLU on the natural course of ERBB2 and Bcl-2 family members was observed in a single the disease. breast cancer cell line, which may not be representative for BNIP3 was the only factor which, at first glance, did not cohorts of patients. However, when we studied three inde- fit the scenario that ERBB2 induces an antiapoptotic pheno- pendent cohorts of node-negative breast cancer patients type because BNIP3 was upregulated in response to ERBB2 (n = 782), all five upregulated antiapoptotic factors induction in MCF-7 cells and it is considered to represent a (MCL-1, TEGT, BAG1, BNIP1, and BECN1)observedin proapoptotic Bcl-2 family member (32–34). However, high NeuT expressing MCF-7 cells showed a correlation with BNIP3 expression was associated with worse prognosis in ERBB2 in at least one of the cohorts. Also, the negative cor- the Mainz and Rotterdam cohorts. Therefore, if the role of relation between ERBB2 and BAX was confirmed. BNIP3 in breast cancer is really proapoptotic, its negative

458 Clin Cancer Res; 16(2) January 15, 2010 Clinical Cancer Research

Bcl-2 Family Members in Breast Cancer

prognostic influence would be counterintuitive. Moreover, we used a murine tumor model that allows the study of genetic and clear biochemical confirmation of BNIP3's pu- ERBB2-dependent tumors (2, 14, 17, 21). Switching off tative role as a BH3-only protein is still awaited. Recently, ERBB2 expression in these tumors leads to apoptosis conflicting results have been reported for immunohisto- and tumor remission in vivo. When we additionally over- chemically determined BNIP3, which has been reported expressed Bcl-xL in the tumor cells, apoptosis and tumor to be associated with good survival in invasive carcinoma remission were clearly reduced in response to the with- but with increased risk of recurrence in ductal carcinoma drawal of ERBB2. This shows that the Bcl-2 family member in situ (35). Our results clearly show that high expression Bcl-xL, which acts similar to MCL-1 and also antagonizes of BNIP3 is associated with worse prognosis and may there- BAX and BMF, could partially substitute for ERBB2 in an- fore be evaluated as a possible therapeutic target. tagonizing apoptosis. It would be helpful to validate this MCL-1, which was upregulated in MCF-7 cells following result of the NIH3T3 cell-based murine tumor model us- the expression of NeuT, was associated with adverse prog- ing epithelial breast cancer cells such as MCF-7. However, nosis in the combined cohort, but not in the individual when we conditionally express ERBB2 in epithelial cells, cohorts. Therefore, confirmation in further studies is re- they activate fail-safe mechanisms such as oncogene-in- quired. All other ERBB2-associated Bcl-2 family members duced senescence or apoptosis (13, 25), which first have were not consistently associated with prognosis. to be bypassed before tumors could be formed. Therefore, It should be considered that the 10 ERBB2-associated systematic in vivo studies on interactions between ERBB2 genes in Table 1 were identified in a single cell line and putative antiapoptotic factors in epithelial cells are (MCF-7). Although the ERBB2 dependence of the 10 genes very difficult to perform. has formally been proven in this particular cell line, it is Our study confirms previous reports that ERBB2 influ- unlikely that the result is representative for populations ences the expression of Bcl-2 family members (6, 12). of patients. Therefore, we systematically studied the rele- Moreover, ERBB2 expression correlated with the expres- vance of all known Bcl-2 family members and related sion of several Bcl-2 family members in patients with genes.7 Analyzing all 28 Bcl-2 family members or associat- breast cancer. However, because of the high interindividu- ed factors that were present on our chip (Supplementary al variability, it would not be reasonable to routinely com- Table S4), we identified only one additional factor consis- bine ERBB2-targeting therapy with a standard Bcl-2 tently associated with prognosis: BCL2L13, which was also family–antagonizing therapy. Rather, patients should be negatively correlated with ERBB2 expression in breast can- tested for overexpression of specific Bcl-2 members to al- cer tissue. High expression of BCL2L13 was associated with low an individualized targeting of the most relevant fac- better prognosis. Interestingly, the prognostic value of tors, such as MCL-1, BMF, BAX, or BNIP3. BCL2L13 was independent from the established clinical factors in the multivariate Cox analysis. BCL2L13 (Bcl- Disclosure of Potential Conflicts of Interest rambo) has not yet been studied in breast cancer. To our knowledge, the only publication reporting a correlation of No potential conflicts of interest were disclosed. BCL2L13 with prognosis in neoplastic diseases is a study in acute lymphoblastic leukemia, in which BCL2L13 was Grant Support associated with L-asparaginase resistance (36). BCL2L13 is a proapoptotic member of the Bcl-2 family (37, 38). Stiftung Rheinland Pfalz for Innovation (Project 646) Therefore, it seems to fit our general pattern that high and by the NGFN-BMBF project OncoProfile. BCL2L13 expression was associated with better prognosis. The costs of publication of this article were defrayed in To functionally validate a role for antiapoptotic Bcl-2 part by the payment of page charges. This article must family members in oncogenic ERBB2 signaling in vivo, therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received 6/26/09; revised 10/29/09; accepted 10/29/09; 7 Corresponding to: http://www-personal.umich.edu/~ino/List/2342.htm. published OnlineFirst 1/12/10.

References 1. Schmidt M, Böhm D, von T, et al. The humoral immune system has a lung cancer is associated with poor prognosis. Int J Cancer 2001;92: key prognostic impact in node-negative breast cancer. Cancer Res 474–9. 2008;68:5405–13. 5. Hengstler JG, Lange J, Kett A, et al. Contribution of c-erbB2 and to- 2. Hausherr CK, Schiffer IB, Gebhard S, et al. Dephosphorylation of p- poisomerase IIα to chemoresistance in ovarian cancer. Cancer Res ERK1/2 in relation to tumor remission after ERBB2 and Raf1 blocking 1999;59:3206–14. therapy in a conditional mouse tumor model. Mol Carcinog 2006;45: 6. Siziopikou KP, Khan S. Correlation of HER2 gene amplification with 302–8. expression of the apoptosis-suppressing genes bcl-2 and bcl-x-L in 3. Tanner B, Hasenclever D, Stern K, et al. ErbB-3 predicts survival in ductal carcinoma in situ of the breast. Appl Immunohistochem Mol ovarian cancer. J Clin Oncol 2006;24:4317–23. Morphol 2005;13:14–8. 4. Micke P, Hengstler JG, Ros R, et al. C-erbB2 expression in small cell 7. Milella M, Trisciuoglio D, Bruno T, et al. Trastuzumab down-regulates

www.aacrjournals.org Clin Cancer Res; 16(2) January 15, 2010 459

Petry et al.

Bcl-2 expression and potentiates apoptosis induction by Bcl-2/Bcl- 26. Rizzi F, Caccamo AE, Belloni L, Bettuzzi S. Clusterin is a short XL bispecific antisense oligonucleotides in ERBB2 gene-amplified half-life, poly-ubiquitinated protein, which controls the fate of breast cancer cells. Clin Cancer Res 2004;10:7747–56. prostate cancer cells. J Cell Physiol 2009;219:314–23. 8. Green DR, Reed JC. Mitochondria and apoptosis. Science 1998;281: 27. Chi KN, Siu LL, Hirte H, et al. A phase I study of OGX-011, a 2′-meth- 1309–12. oxyethyl phosphorothioate antisense to clusterin, in combination 9. Cory S, Adams JM. The Bcl2 family: regulators of the cellular life-or- with docetaxel in patients with advanced cancer. Clin Cancer Res death switch. Nat Rev Cancer 2002;2:647–56. 2008;14:833–9. 10. Gee JM, Robertson JF, Ellis IO, et al. Immunocytochemical localiza- 28. Toffanin S, Daidone MG, Miodini P, et al. Clusterin: a potential target tion of BCL-2 protein in human breast cancers and its relationship to for improving response to antiestrogens. Int J Oncol 2008;33:791–8. a series of prognostic markers and response to endocrine therapy. 29. Shannan B, Seifert M, Boothman DA, Tilgen W, Reichrath J. Clusterin Int J Cancer 1994;59:619–28. over-expression modulates proapoptotic and antiproliferative effects 11. Leek RD, Kaklamanis L, Pezzella F, Gatter KC, Harris AL. Bcl-2 in of 1,25(OH)2D3 in prostate cancer cells in vitro. J Steroid Biochem normal human breast and carcinoma, association with oestrogen re- Mol Biol 2007;103:721–5. ceptor-positive, epidermal growth factor receptor-negative tumours 30. KrügerS,OlaV,FellerAC,FischerD,FriedrichM.Expressionof and in situ cancer. Br J Cancer 1994;69:135–9. cancer-testis antigen CT7 (MAGE-C1) in breast cancer: an immuno- 12. HensonES,HuX,GibsonSB.HerceptinsensitizesErbB2- histochemical study with emphasis on prognostic utility. Pathol overexpressing cells to apoptosis by reducing antiapoptotic Oncol Res 2007;13:91–6. Mcl-1 expression. Clin Cancer Res 2006;12:845–53. 31. Zhang S, Zhang D, Zhu Y, Guo H, Zhao X, Sun B. Clusterin expres- 13. Trost TM, Lausch EU, Fees SA, et al. Premature senescence is a sion and univariate analysis of overall survival in human breast primary fail-safe mechanism of ERBB2-driven tumorigenesis in cancer. Technol Cancer Res Treat 2006;5:573–8. breast carcinoma cells. Cancer Res 2005;65:840–9. 32. Kubasiak LA, Hernandez OM, Bishopric NH, Webster KA. Hypoxia 14. Schiffer IB, Gebhard S, Heimerdinger CK, et al. Switching off ERBB2/ and acidosis activate cardiac myocyte death through the Bcl-2 fam- neu in a tetracycline controlled mouse tumor model leads to apoptosis ily protein BNIP3. Proc Natl Acad Sci U S A 2002;99:12825–30. and tumor-size dependent remission. Cancer Res 2003;63:7221–31. 33. Sowter HM, Ratcliffe PJ, Watson P, Greenberg AH, Harris AL. HIF-1- 15. Gossen M, Freundlieb S, Bender G, Müller G, Hillen W, Bujard H. dependent regulation of hypoxic induction of the cell death factors Transcriptional activation by tetracyclines in mammalian cells. Sci- BNIP3 and NIX in human tumors. Cancer Res 2001;61:6669–73. ence 1995;268:1766–9. 34. Boyd JM, Malstrom S, Subramanian T, et al. Adenovirus E1B 19 kDa 16. Gossen M, Bujard H. Tight control of gene expression in mammalian and Bcl-2 proteins interact with a common set of cellular proteins. cells by tetracycline-responsive promoters. Proc Natl Acad Sci U S A Cell 1994;79:341–51. 1992;89:5547–51. 35. Tan EY, Campo L, Han C, et al. BNIP3 as a progression marker in 17. Baasner S, Melchner H, Klenner T, Hilgard P, Beckers T. Reversible primary human breast cancer; opposing functions in in situ versus tumorigenesis in mice by conditional expression of the HER2/ invasive cancer. Clin Cancer Res 2007;13:467–74. c-erbB2 receptor tyrosine kinase. Oncogene 1996;13:901–11. 36. Holleman A, den Boer ML, de Menezes RX, et al. The expression of 18. Yamamoto S, Nagao M, Sugimori M, et al. Transcription factor ex- 70 apoptosis genes in relation to lineage, genetic subtype, cellular pression and Notch-dependent regulation of neural progenitors in drug resistance, and outcome in childhood acute lymphoblastic the adult rat spinal cord. J Neurosci 2001;21:9814–23. leukemia. Blood 2006;107:769–76. 19. Huber C, Bobek N, Kuball J, et al. Inhibitors of apoptosis confer re- 37. Perier C, Tieu K, Guegan C, et al. Complex I deficiency primes sistance to tumour suppression by adoptively transplanted cytotoxic Bax-dependent neuronal apoptosis through mitochondrial oxidative T-lymphocytes in vitro and in vivo. Cell Death Differ 2005;12:317–25. damage. Proc Natl Acad Sci U S A 2005;102:19126–31. 20. Wesarg E, Hoffarth S, Wiewrodt R, et al. Targeting BCL-2 family 38. Puthalakath H, Villunger A, O'Reilly LA, et al. Bmf: a proapopto- proteins to overcome drug resistance in non-small cell lung cancer. tic BH3-only protein regulated by interaction with the myosin V Int J Cancer 2007;121:2387–94. actin motor complex, activated by anoikis. Science 2001;293: 21. Hermes M, Schormann W, Brulport M, et al. Trastuzumab therapy vs 1829–32. tetracycline controlled ERBB2 downregulation: influence on tumour 39. Boise LH, Gonzalez-Garcia M, Postema CE, et al. Bcl-x, a bcl-2- development in an ERBB2-dependent mouse tumour model. Br J related gene that functions as a dominant regulator of apoptotic Cancer 2008;98:1525–32. cell death. Cell 1993;74:597–608. 22. Wang Y, Klijn JG, Zhang Y, et al. Gene-expression profiles to predict 40. Marzo I, Brenner C, Zamzami N, et al. Bax and adenine nucleotide distant metastasis of lymph-node-negative primary breast cancer. translocator cooperate in the mitochondrial control of apoptosis. Sci- Lancet 2005;365:671–9. ence 1998;281:2027–31. 23. Desmedt C, Piette F, Loi S, et al. Strong time dependence of the 41. Qin W, Hu J, Guo M, et al. BNIPL-2, a novel homologue of BNIP-2, 76-gene prognostic signature for node-negative breast cancer interacts with Bcl-2 and Cdc42GAP in apoptosis. Biochem Biophys patients in the TRANSBIG multicenter independent validation series. Res Commun 2003;308:379–85. Clin Can Res 2007;3:3207–14. 42. Reimers K, Choi CY, Bucan V, Vogt PM. The Bax Inhibitor-1 (BI-1) 24. Loi S, Haibe-Kains B, Desmedt C, et al. Definition of clinically distinct family in apoptosis and tumorigenesis. Curr Mol Med 2008;8:148–56. molecular subtypes in estrogen receptor-positive breast carcinomas 43. Götz R, Wiese S, Takayama S, et al. Bag1 is essential for differenti- through genomic grade. J Clin Oncol 2007;25:1239–46. ation and survival of hematopoietic and neuronal cells. Nat Neurosci 25. Spangenberg C, Lausch EU, Trost TM, et al. ERBB2-mediated tran- 2005;8:1169–78. scriptional up-regulation of the α5β1 integrin fibronectin receptor 44. Takayama S, Sato T, Krajewski S, et al. Cloning and functional promotes tumor cell survival under adverse conditions. Cancer Res analysis of BAG-1: a novel Bcl-2-binding protein with anti-cell 2006;66:3715–25. death activity. Cell 1995;80:279–84.

460 Clin Cancer Res; 16(2) January 15, 2010 Clinical Cancer Research

ERBB2 Induces an Antiapoptotic Expression Pattern of Bcl-2 Family Members in Node-Negative Breast Cancer

Ilka Brigitte Petry, Esther Fieber, Marcus Schmidt, et al.

Clin Cancer Res Published OnlineFirst January 12, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-09-1617

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2010/01/18/1078-0432.CCR-09-1617.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2010/01/10/1078-0432.CCR-09-1617. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.